On 06/09/2024 18:18, Niklas Söderlund wrote:
On 2024-09-06 17:13:28 +0300, Tomi Valkeinen wrote:
Awesome! I will try to find some time soon and rebase that series, I
think I have an idea on how to avoid having to mock around in the
v4l-async core and still achieve the same thing.
Hmm, do you mean "[PATCH] media: v4l: async: Fix completion of chained
subnotifiers"? I only now noticed that. I did not apply that... Should I
have?
If you got video nodes from rcar-vin then there is no need. Without it
there is a dependency on probe order for the VIN instances, if done in
"incorrect" order the complete callback is not called and no video
device nodes are registered.
I noticed that if I don't load the max96712 driver at all, I still get a
media pipeline, with only rcar_isp and rcar_csi2 in it. I'd expect to
get no media pipeline as a component is missing. I applied the patch
above, but it didn't affect this problem.
Tomi
